False positives · student, any assignment, first time it happens
Your essay got flagged as AI. Here is the first hour.
Somewhere between the moment a score shows up on a rubric and the moment you actually talk to your instructor, there is a gap where nothing has happened yet. That gap is where most of the damage gets done, because it is also where the panic is worst. This page is for that gap: the hour after you find out, before you have written anything to anyone.
A detector flagged your essay, and right now that is a number, not a verdict. Nothing has been decided, no one has ruled on anything, and the score alone does not prove what you did or didn't do. The mistake most students make in the first hour is treating the flag as an accusation and reacting to it like one, firing off a defensive message before they know what was actually raised or by whom. Read what the tool told your instructor, if you can see it. Find out whether this is a flag or a formal charge. Gather your drafts before you do anything else. Then, and only then, decide what to say.
Why this happens
AI detectors score text on how predictable each word is given the words before it. Machine-generated text tends to pick the statistically likely next word more often than a person does, so low unpredictability correlates with machine authorship, on average, across the texts the tool was trained on. It is a correlation, not a fingerprint. Essays written to a rubric, in a second language, or in a hurry after outlining carefully all compress a writer's natural variation, which is the exact thing the score is measuring. None of that means the software knows who typed the words. It means the score and the truth can point in different directions, and an instructor who trusts the number without knowing that will read it as more certain than it is.
Where honest writing and machine writing genuinely overlap
These are real patterns, not coincidences. Each one is something a careful human writer in this situation has good reason to do, and something a model does for entirely different reasons.
- Formulaic five-paragraph structure
- Thesis, three body points, restated conclusion, taught explicitly in most US high schools, produces the same predictable sentence-to-sentence transitions a detector is built to catch. The structure is a taught skill, not evidence of a machine.
- Low sentence-length variation
- Writers who outline carefully and then fill in each point in order tend to produce sentences of similar length and shape all the way through, because the planning already did the varying work that improvisation would otherwise add.
- Topic-sentence-first paragraphing
- Academic writing courses drill 'state your point, then support it' into every paragraph. That is also the shape a detector's training data associates most strongly with generated text, since it optimizes for the same predictability.
- Careful hedge language
- Students told to 'avoid overstating your claim' add qualifiers like 'suggests' and 'may indicate' throughout an essay. Detectors read low certainty language as a flatness signal, the same flatness a cautious, well-trained writer produces on purpose.
What actually helps
Before you write a single word back, collect what you already have: the document's version history if it was written in Google Docs or Word online, any earlier drafts saved locally or emailed to yourself, notes, outlines, or a marked-up source you annotated. Check whether your school's policy treats a detector flag as a preliminary screen or as evidence on its own; the two call for different responses, and you won't know which one applies to you until you actually find and read the policy. Ask your instructor, calmly and in writing, what specifically was flagged, rather than denying the whole essay. A conversation about three sentences is easier to have than a conversation about your integrity.
What the flattening looks like
On the left, honest writing with the specifics drained out of it, which is the shape that reads as machine-written. On the right, the same content with the writer's own detail restored. The voice is The Perfectionist
In today's fast-paced world, the theme of isolation stands as a testament to the human condition explored throughout the novel. It is not merely a plot device, but a profound reflection of society's ever-evolving landscape. Furthermore, the author's use of symbolism serves as a testament to this central idea, and it is important to note that the imagery reinforces this theme throughout.
The novel uses isolation not as a plot device, though one could call it that, but as a kind of mirror held up to something we already know but seldom say aloud. It is a quiet recognition, perhaps: that the condition of being alone is not the exception but the rule we keep trying to outrun. The imagery does not merely illustrate this; it insists on it. Every symbol returns, with slight variation, to the same refusal of connection. One begins to wonder whether the author chose the symbols or the symbols chose the author.
Steps worth taking
- Separate the score from the accusation A flag is a number attached to your document. An accusation is a person deciding to act on that number. Find out which stage you are actually in before you respond to either.
- Pull your process evidence together first Version history, saved drafts, browser search history for sources, annotated PDFs. Gather it before you talk to anyone, not after, so you are not scrambling for it mid-conversation.
- Ask what specifically was flagged Request the passages, not just the score. Most detectors highlight sections rather than condemning the whole document, and you can only respond to something concrete.
- Write the first message slow A calm, specific message asking for the flagged passages and your school's process lands better than a long defense of your character. Save the defense for if you actually need one.
Questions
Why is my writing being flagged as AI when I wrote it myself?
Detectors score predictability, not authorship. Formulaic structure, careful hedging, and low sentence-length variation, all things honest academic writing produces on purpose, score close to what generated text produces. The flag means your text looked statistically machine-like to that tool. It does not mean the tool knows who wrote it.
What should I do first if my paper is flagged?
Find out whether it is a preliminary flag or a formal charge, and gather your drafts, version history, and notes before you respond to anyone. Order matters here more than speed.
What happens if my college essay gets flagged for AI?
It depends entirely on your school's policy, which is why finding that policy is the second thing to do, right after separating the score from an accusation. Some schools treat a detector flag as inconclusive on its own; others require it be paired with other evidence before any charge is brought.
Should I just deny it happened?
Denial without specifics is the weakest response available, because it gives the other person nothing to check against. Naming your drafting process and offering to show your work moves the conversation further than insisting you didn't do it.
What does underslop actually have to do with this situation?
underslop edits an AI draft so it reads like the person who wrote the voice settings, for someone starting from a machine draft honestly. If your essay is entirely your own writing, this page is not about the tool. It exists to help you handle the conversation in front of you.
underslop edits drafts for voice. It does not check, score, or contest anything, and nothing on this page will change a result you have already been given. What it can do is show you which patterns in a piece of writing read as machine-written, using the same open lint the rest of the site runs on.